Composition and Biological Properties of Lipopolysaccharides Isolated from Schizothrix Calcicola (Ag.) Gomont (Cyanobacteria) GEORG KELETI,'* JAN L

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Sept. 1979, p. 471-477 Vol. 38, No. 3 0099-2240/79/09-0471/07$02.00/0

Composition and Biological Properties of Lipopolysaccharides Isolated from Schizothrix calcicola (Ag.) Gomont (Cyanobacteria) GEORG KELETI,'* JAN L. SYKORA,' EDWIN C. LIPPY,2 AND MAURICE A. SHAPIRO' Graduate School ofPublic Health, University ofPittsburgh, Pittsburgh, Pennsylvania 15261,' and U.S. Environmental Protection Agency, Health Effects Research Laboratory, Cincinnati, Ohio 452682 Received for publication 15 March 1979 The most common cyanobacterium contaminating drinking water systems in southwestern Pennsylvania is Schizothrix calcicola. Lipopolysaccharides (LPS) were isolated from this species by hot phenol-water extraction. The polysaccharide moiety was composed of glucosamine, galactose, glucose, mannose, xylose, and rhamnose. The lipid A part contained beta-hydroxylauric, myristic, pentadecanoic, palmitic, beta-hydroxypalmitic, stearic, oleic, and linoleic acids. In contrast to many LPS isolated from Enterobacteriaceae, the dominant component was not beta-hydroxymyristic but beta-hydroxypalnitic acid. The LPS induced Limulus lysate gelation and Schwartzman reaction but was nontoxic to mice. The identity of LPS was verified by alkali and lysozyme treatment. The results suggest that S. calcicola is one of the principal sources of endotoxins in water systems using open finished-water reservoirs. A water-bome outbreak of gastroenteritis af- [LPS]) have been isolated from six species of fected approximately 5,000 persons in Sewickley, cyanobacteria (2, 17, 29, 30). LPS are constitu- Pa., during August 1975. Extensive microbiolog- ents of the outer cell wall of some microorga- ical and chemical analyses of specimens ob- nisms, predominantly gram-negative bacteria. tained from patients and of water samples failed These macromolecular substances are composed to identify a causative agent. However, investi- ofthree regions. The first region is the 0-specific gation of the Sewickley water system revealed polysaccharide consisting of repeating oligosac- an accumulation of Schizothrix calcicola (Cy- charide units, different in various species of bac- anobacteria = Cyanophyta) in the open fin- teria. The second region is the basal core oligo- ished-water reservoirs (15). Even 1 month after saccharide, and the third region is the hydro- the outbreak, the water in the reservoir with the phobic lipid part, the lipid A. Many gram-nega- longest detention time was still contaminated by tive bacterial mutants possess only glycolipids, this species (400,000 cells ml-'). and the 0-specific polysaccharide is absent. Toxic algae and cyanobacteria have been rec- Sometimes even a part of the core oligosaccha- ognized since antiquity. They are found in di- ride is lacking. verse aquatic environments and have been im- The LPS derived from blue-green organisms plicated in poisoning and death of fish, wildlife, are generally identical to those occurring in domestic animals, and humans (18). Many ob- gram-negative bacteria. However, the available served episodes were caused by ingestion of wa- though limited qualitative data on the compo- ter from reservoirs, lakes, and ponds contamined sition of cyanobacterial LPS have already re- by cyanobacteria. Since these organisms are vealed a diversity in their chemical composition widely distributed, it is not surprising that cy- and biological characteristics. anobacteria were associated with several water- In this report we describe the composition and borne outbreaks of gastroenteritis in the United biological properties of LPS isolated from S. States (25, 26), the Philippines (3), and India (7). calcicola. S. calcicola is not usually listed among the toxic cyanobacteria, and there is only one publication MATERIALS AND METHODS suggesting that this species may form toxic Enumeration of algae and cyanobacteria. Al- strains (1). gae and cyanobacteria were counted in samples col- In addition to toxins (anatoxin, saxitoxin, toxic lected from five different water systems in the Pitts- polypeptides), macromolecular substances char- burgh, Pa., area. All of the water treatment plants acterized as endotoxins (lipopolysaccharides selected for sampling stored finished water in open 471 472 KELETI ET AL. APPL. ENVIRON. MICROBIOL. reservoirs which varied in size from 0.5 to 177.5 million eliminated by enzymatic hydrolysis with a beta-glu- gallons. The samples were collected at weekly inter- cosidase as recommended by Volk (27). Cellulase (2.5 vals during late spring and early summer (April-June mg ml-'; Sigma Chemical Co.) was dissolved in water 1977) from the treated water in each plant, open and centrifuged at 37,000 x g for 15 min, and the reservoir(s), and distribution system(s). Algae were supernatant was passed through a 0.45-nm-pore size preserved with Lugol solution, concentrated by sedi- membrane filter (Millipore Corp.). The crude LPS was mentation, and enumerated in a Palmer-Maloney nan- suspended at 5 mg ml-' in 0.05 M citrate buffer (pH noplankton cell. The results were expressed as number 4), and a final concentration of 0.25 mg of enzyme was of cells per milliliter. added per 1 mg of LPS. After 10 h of treatment at Cultures of S. calcicola. A brown strain of S. room temperature, a second equal amount of cellulase calcicola identical to that occurring in drinking water was added, and the enzymatic hydrolysis was contin- was isolated from Mosquito Creek Reservoir, near ued overnight. Afterwards the mixture was dialyzed Youngstown, Ohio. It was identified by using cell for 48 h against tap water, concentrated in a rotary morphology, color, and size and by comparison with evaporator, and freeze-dried. To remove most of the the organisms obtained from Sewickley Reservoir no. nucleic acids, an aqueous solution (1 to 2%) of the 4. The identification was confirmed by Francis Drouet freeze-dried, crude LPS was further purified by re- of the Academy of Natural Sciences of Philadelphia. peated (three times) centrifugation at 105,000 x g in S. calcicola was mass cultured in 9-liter bottles using a Spinco ultracentrifuge for 4 h. The sediment was Allen medium (9) kept at 24°C, aerated with com- resuspended in distilled water and centrifuged for 10 pressed air at a rate of 100 to 250 ml/min, and illumi- min at 3,000 rpm. The supernatant, representing the nated by white fluorescent light adjusted to a 16-h purified LPS, was lyophilized. cycle. Each bottle was inoculated with 200 ml of pure, The phenol phase of the extracted cells was sub- aerated starter culture. For comparison, several mass jected to the same procedure as the aqueous phase, cultures of S. calcicola were prepared in a 14-liter and only a negligible contamination by LPS was iso- Microferm MF-114 fermentor (New Brunswick Sci- lated from this fraction. entific Corp.) with aeration and illumination and Analytical procedures. The neutral and amino equipped with constant temperature controls. For sugars were released from the LPS by using hydrolysis endotoxin isolation, 14-day-old cultures were har- with 1 N H2SO4 and 4 N HCl in a boiling-water bath vested at 3,450 rpm using a flow-through centrifuge for 4 and 12 h, respectively. After neutralization and (Gyrostester, DeLaval Separator Co.), washed three subsequent centrifugation at 2,000 rpm for 10 min at times, and lyophilized. 4°C, the supernatant was cooled and dried in a vacuum The heterotrophic bacterial contamination of S. desiccator equipped with P205 and, in the case of calcicola was determined by plating serial dilutions amino sugars, also with NaOH pellets. Final analysis (0.85% saline) of the cultures on Trypticase soy agar of the released sugars was performed using descending (BBL Microbiological Systems) in duplicate. The paper chromatography with proper sugar standards plates were incubated for 72 h at 20 and 35°C. Because on Whatman no. 1 paper using n-butanol-pyridine- the majority ofthe bacteria contaminating S. calcicola water (6:4:3, vol/vol/vol) (12). were from the family Enterobacteriaceae, a laboratory For fatty acid analysis of the lipid A part, the LPS strain of Salmonella typhimurium was selected as the was hydrolyzed with 4 N HCl in a boiling-water bath biomass indicator. The bacterial biomass was calcu- for 5 h, followed by chloroform extraction repeated lated utilizing a 24-h culture of this strain in soybean three times. The combined chloroform extract was casein digest broth as a standard. A series of dilutions dried in a vacuum rotary evaporator, and the residue in physiological saline containing this organism was was dissolved in ethyl ether and dried in a sand bath plated on Trypticase soy agar medium, and the num- at 37°C. The sample was methylated with 0.5 ml of ber of colonies of S. typhimurium present in the cul- BF3 methanol (14%, wt/vol) in a tightly stoppered ture was determined. One liter of the same culture was flask, placed in a boiling-water bath for 2 min, and, centrifuged, washed three times, and lyophilized, and after cooling, extracted twice with petroleum ether. the dry weight of bacteria was established. Based on The combined petroleum ether phases were dried at this standard, the average count of heterotrophic bac- 60°C in a sand bath and dissolved in ethyl acetate teria contaminating S. calcicola cultures constituted before chromatographic and mass spectrometric anal- only 0.04% of the total biomass. ysis. Isolation and purification ofLPS. The harvested The methyl esters of the fatty acids were examined cells were washed three times with sterile, distilled on a Packard series 7400 gas chromatograph (Packard water. Each washing in a Sorvall (RC-2B) centrifuge Instrument Co., Downers Grove, Ill.) using a 3% OV- lasted 25 min at 5,500 rpm and 4°C. The resultant 17 glass column (2 mm by 10 feet [ca. 3.05 ni]). The sediment was uniformly suspended in distilled water analyses were programmed from 100 to 300°C at 7°C/ and freeze-dried. The lyophilized, LPS-containing ma- min and held for 15 min with a chart speed of 0.25 terial was extracted twice for 30 min with 45% phenol inch (ca. 6.4 mm) per min. Helium was used as a solution at 68°C according to Westphal and Jann (31). carrier gas. The single peaks were further analyzed by The aqueous phase was dialyzed against distilled water using a LKB 900 mass spectrometer. until all traces of phenol were removed, and then it The following determinations were made by the was centrifuged at 5,000 rpm for 15 min. The super- respective methods: 2-keto-3-deoxyoctonate (KDO) natant was freeze-dried. by the thiobarbituric acid method; heptoses by the It has been shown that crude LPS from cyanobac- technique of Osborne; carbohydrates by the phenol- teria may contain glucan (17, 29). This compound was sulfuric acid method; proteins and total phosphorus VOL. 38, 1979 SCHIZOTHRIX CALCICOLA LIPOPOLYSACCHARIDES 473 by the techniques of Lowry et al. All these chemical The most common cyanobacterium in open fin- analyses were performed as microassays as described ished-water storage facilities in the Pittsburgh in the Keleti and Lederer handbook (12). In addition, area (and probably in the entire United States) the KDO was determined by the improved microassay in three technique described by Karkhanis et al. (10) using is S. calcicola. It was frequently found hydrolysis by 0.2 N H2SO4 at 100°C for 30 min. out of six reservoirs and one distribution system. Biological activity. The lethal effect of the puri- However, the cellular concentrations were usu- fied endotoxin was tested in duplicate by intraperito- ally low, with high values recorded on few oc- neal injections into five 20-g Swiss-Webster female casions (Table 1). Lesser quantities of algae and mice for each concentration (100 ag, 500 ,tg, 1 mg, 2 cyanobacteria were present in distribution sys- mg, 4 mg). tems, indicating contamination ofdrinking water Schwartzman phenomenon was performed on New during storage (Table 1). A portion of the pho- Zealand rabbits using three different concentrations of tosynthetic organisms was destroyed by residual LPS (80, 40, and 20 ILg). The LPS in phosphate- darkness or pres- buffered, pyrogen-free saline (pH 7.2) was injected chlorine or by the permanent subcutaneously into two adult rabbits. Sterile phos- sure in the mains. However, even in the adverse phate-buffered saline (pH 7) served as a control. environment of distribution systems the living Twenty-four hours later a challenge injection of LPS cells may persist for a prolonged period, because (40 ,ug) was given intravenously. The rabbits were not all treatment plants postchlorinate the water observed 24 h later for signs ofhemorrhage or necrosis. leaving the reservoirs and entering the distribu- LAL test. For the Limulus amoebocyte lysate tion system. The chlorine content was very low (LAL) test, several serial solutions of LPS were pre- or at times absent in tap water, with means pared in pyrogen-free water (Travenol for injection ranging from 0 to 0.6 mg of Cl per ml. [Travenol Lab. Inc.], divided into 100-ml portions and LPS isolated from S. calcicola contained autoclaved for 1 h in 250-ml flasks). The assays were The performed by adding 0.1 ml of the reconstituted lysate the neutral sugars glucose, galactose, mannose, each to pyrogen-free tubes (10 by 75 mm), followed by xylose, and rhamnose, whereas the amino sugars the addition of 0.1 ml of LPS suspension. The tubes were represented only by glucosamine (Table 2). were thoroughly agitated by a Vortex mixer, covered Both assays for KDO, including an improved with aluminum foil, and incubated for 60 min at 37°C microassay technique (10), were negative, and in a water bath. A positive control containing 2.5 ng no color was obtained, indicating that this sub- ml-' of Escherichia coli endotoxin and a negative stance was not present in the LPS of S. calcicola control (pyrogen-free water) were used in the assay. (Table 2). The absence of KDO is also indicative After the incubation period, the tubes were carefully of insignificant bacterial contamination, because inverted to determine presence of gelation. is a well-known constituent of Lysozyme and alkali treatment. The macromo- this component lecular compounds isolated from S. calcicola were LPS of gram-negative bacteria. KDO was also tested for LPS content and possible occurrence of absent in LPS derived from Anabaena varia- peptidoglycan by using lysozyme and alkali treatment. bilis (29). Buttke and Ingram (2) found only a The isolated material was treated with three-times- negligible quantity (1.13% of LPS) of this com- crystallized egg white lysozyme (100lOg mg-') at pH pound in Agmnenellum quadruplicatum, 6.2 in 0.85% saline kept in a water bath at 37°C for 18 whereas LPS from Synechococcus (Anacystis) h (32). nidulans and Phormidium spp. contained from Alkali treatment was accomplished using the tech- 0.5 to 1.5% (17, 30). nique described by Suzuki et al. (24). The material Similarly, as in the other cyanobacteria inves- was suspended in pyrogen-free 0.1 N sodium hydroxide in solution, which was kept in a sealed tube in a water tigated for LPS content, heptose was absent bath at 700C for 4 h. After neutralization with pyrogen- S. calcicola LPS. However, aminoheptose was free hydrochloric acid, the samples were subjected to discovered by Weise et al. in Synechococcus the LAL gelation test. (Anacystis) nidulans LPS. S. calcicola LPS contained 63% carbohy- RESULTS drates, whereas Phormidium spp., Agmenellum The field study results show the absence of quadruplicatum, Anabaena variabilis, and Sy- living photosynthetic organisms (algae and cy- nechococcus (Anacystis) nidulans LPS con- anobacteria) in plant-finished water. However, tained 60, 59.5, 80.3, and 60%, respectively (2, 17, substantial accumulations of these nuisance or- 29, 30). ganisms were found in open finished-water res- The total phosphorus content in cyanobacte- ervoirs. rium LPS was found to be rather low, ranging Green algae (Chlorophyta) are the most com- from less than 1% to 3%, whereas in S. calcicola mon and abundant nuisance organisms contam- LPS it constituted less than 0.1%. inating storage reservoirs and distribution sys- The LPS from S. calcicola was contaminated tems during the summer. Cyanobacteria are by 7.8% proteins. Phormidium spp., Anabaena quantitatively less important but may develop variabilis, and Agmenellum quadruplicatum into a monospecific bloom if left unchecked (15). LPS contained 7 to 20, 8.4, and 0.13% proteins, 474 KELETI ET AL. APPL. ENVIRON. MICROBIOL. TABLE 1. Distribution of algae and cyanobacteria in five drinking water systems located in Pittsburgh, Pa., area Total autrophic organisms Total cyanobac- Schizothrix calcicola (cells mV1') teria (cells ml-') Fre- Locality Na MeanMean Range quency of Range occur- Mean Range Mean Range (cells (cells rence (% mE)fl 1) of samples) Dixmont Reservoir 10 77,400 2,400-216,600 160 0-1,500 160 0-1,500 20 Dixmont distribution 8 34,800 520-96,800 19 0-150 19 0-150 13 Highland Park I Reservoir 12 5,800 360-23,800 56 0-250 56 0-250 42 Highland Park (Pitts- 9 3,500 340-12,000 53 0-220 0 0 0 burgh) distribution McKeesport Reservoir 12 2,100 110-6,500 470 0-2,100 460 0-2,100 67 McKeesport distribution 9 4,900 190-32,900 320 0-2,000 180 0-1,000 44 Sewickley no. 3 Reservoir 10 530 45-1,800 12 0-120 12 0-120 30 Sewickley no. 4 Reservoir 9 17,100 20-52,210 140 0-1,200 140 0-1,200 56 Sewickley low-pressure 4 270 150-300 0 0 0 0 0 distribution Sewickley high-pressure 4 1,700 300-5,200 0 0 0 0 0 distribution Springdale Reservoir 11 286,700 133,800-490,300 290 0-2,300 280 0-2,100 18 Springdale distribution 9 5,100 2,800-16,690 53 0-360 0 0 0 aN, Sample size. TABLE 2. Sugars, phosphorus, and proteins occurring in LPS of cyanobacteria Sugars

Phos- Pro- SpeciesSpecies ~~~~~~~~~~~~~~~~~KDO() Tarotaldae pho%rus teinr ~~(%) ~drates

Agmenellum quadru- + + + - - + - + NDXb+ - 0.13 2.9 0.13 plicatum' Anabaena variabilisc + + + - - - - + + - _ + - 0 80.3 0.03 8.4 Anacystis nidulansd + + - - - - ++ - + + 1.5 60 <1 ND Phormidium spp.e + + + - + + - + - + + + - 0.5-1.0 60 <1 7.20 Schizothrix calcicola + + + - - + - + - - - + - 0 63 <0.1 7.8 a Data from Buttke and Ingram (2). b ND, Not done. 'Data from Weckesser et al. (29). d Data from Weise et al. (30). 'Data from Mikheyskaya et al. (17). f ?, Unidentified. respectively (2, 17, 29). species of Phormidium (17), a genus closely Gas-liquid chromatography and mass spec- related to Schizothrix. Several species of Phor- trometry identified eight long-chain fatty acids: midium are even thought to be ecophenes of S. beta-hydroxylauric, myristic, pentadecanoic, calcicola (5). palmitic, beta-hydroxypalmitic, stearic, oleic, The isolated and purified LPS was nontoxic and linoleic. Stearic, oleic, and linoleic acids, to mice when injected intraperitoneally even at containing 18 carbons each, were identified on concentrations as high as 4,000 ,ug per mouse. the basis of differences in their molecular Similarly, Weise et al. (30) have shown that LPS weights as detected by mass spectrometry and isolated from the cyanobacterium Synechococ- by comparison with proper standards. Surpris- cus (Anacystis) nidulans was not toxic to mice. ingly, the dominant component of lipid A was Mikheyskaya et al. (17) have demonstrated that beta-hydroxypalmitic acid (Table 3), a constit- glucan-free LPS from Phormidium spp. did not uent previously shown to be absent from several show any toxicity to mice either. VOL. 38, 1979 SCHIZOTHRIX CALCICOLA LIPOPOLYSACCHARIDES 475 TABLE 3. Fatty acids in LPS in cyanobacteria Agmenellum Anabaena Phormidium Schizothrix Fatty acid quadruplica- Anacystis turn" variabilisb nidulans' app.d calcicola Beta-hydroxylauric - - - - + Beta-hydroxymyristic + + + Beta-hydroxypalmitic +?e + - - + Beta-hydroxystearic +? + Behenic + - + - Lauric - - + - - Linoleic - - - - + Myristic - - - - + Oleic - - + + + Palmitic + + + + + Pentadecanoic - - - _ + Stearic - - + + + Unidentified - + aData from Buttke and Ingram (2). b Data from Weckesser et al. (29). c Data from Weise et al. (30). d Data from Mikheyskaya et al. (17). e ?, Tentative identification. Positive Schwartzman reaction using LPS iso- TABLE 4. LAL gelation of LPS from S. calcicola lated from S. calcicola was observed at concen- Gelation' trations of 40 and 80 ,ug. Signs of hemorrhaging LPS concn (jug and necrosis were observed on the sites of LPS m1-') No pre- Lysozyme NaOH injections 24 h later. Buttke and Ingram (2) treatment treatment treatment recorded the same phenomenon when the LPS 500 +++ +++ + +++ +++ + isolated from Agmenellum quadruplicatum was 100 tested. 50 +++ ND ND ++ ++ - The result of the LAL reaction was positive 10 1 + ++ _ for S. calcicola LPS (Table 4). The biological 0.5 + + - testing also included LAL assays with lysozyme- and alkali-treated LPS. The gelation almost dis- "+++, Firm clot; ++, clot breaks; +, increase in appeared after the treatment with alkali, viscosity; -, no reaction; ND, not done. whereas the lysozyme left the LPS intact. These results suggest that the purified phenol extract unusually high (7%). This suggested contami- from S. calcicola contains true LPS and that nation of the LPS by glucan, a phenomenon the LAL gelation is not caused by peptidogly- noted by several authors (6, 17, 29). An electron cans or other compounds which are also respon- microscopy study of cyanobacteria detected mu- sible for a positive LAL reaction (24, 33). cous cellular envelopes containing glucan (6). DISCUSSION Weckesser et al. (29) found the same substance The planktonic and benthic strains of S. cal- in the water phase of a phenol-water extract ciola have been found in a wide variety of envi- from Anabaena variabilis, whereas LPS was ronmental conditions, and this species was de- isolated only from the phenol phase. In addition, scribed as "perhaps the most widely distributed glucans were also discovered in crude LPS iso- and common blue-green alga on earth" (5). It lated from the cell walls of the genus Phormi- has been discovered in hot springs, in meltwaters dium (17). After the removal of glucan, the of Antarctica, in the Dead Sea at an elevation of average yield of purified LPS from whole S. -1,286 feet (ca. -391.9 m), and in the Himalayas calcicola cells decreased to 1.5 to 2%, which is at an altitude of over 17,000 feet (ca. 5,182 mi). It similar to gram-negative bacteria. is not surprising that S. calcicola can withstand The chemical analysis of LPS from S. calci- adverse conditions in open finished-water res- cola revealed that KDO was not present in this ervoirs and can form occasional water blooms in macromolecular substance (Table 2). However, drinking-water systems. Therefore, the possible this component was absent in endotoxins iso- health effect of LPS derived from this species is lated from Anabaena variabilis (29) and was worthy of investigation. found in small or negligible quantities in endo- The yield of LPS isolated from S. calcicola as toxins ofthree other species [Agmenellum quad- compared to other cyanobacteria (17, 29, 30) was ruplicatum, Synechococcus (Anacystis) nidu- 476 KELETI ET AL. APPL. ENVIRON. MICROBIOL. lans, Phormidium spp.] (2, 17, 30). This com- pressed patients and infants. Enhanced gastroin- pound is lacking even in some gram-negative testinal permeability in infants suggests that bacteria such as Bordetella (16). Therefore, absorption of exogenously derived endotoxin KDO is not necessarily an essential component (from milk or water) could initiate illness under of the core part of LPS. certain conditions. For instance, interaction of Another important aspect is the possible use lead or some other agents present in the envi- of the chemical composition of LPS for taxo- ronment (i.e., water) with endotoxin may result nomical classification of cyanobacteria. In the in harmful effects upon oral ingestion (4). Inter- past, similar characterization has been per- estingly enough, the highest value (800 ng ml-') formed on Salmonella and Escherichia coli, of LPS was detected by DiLuzio and Friedman when serotypes were classified according to the (4) in the tap water of Mexico City. This high sugar composition of their LPS into 40 groups endotoxin level does not seem to affect the in- called chemotypes (11, 19). Similarly, the Citro- digenous population but may cause "traveler's bacter serotypes were divided into 20 chemo- diarrhea" in tourists. types by Keleti et al. (13, 14) and Sedlak et al. The presence of endotoxins in drinking water (21, 22). and the effect of LPS introduced into the blood- Until recently the taxonomy of cyanobacteria stream have been observed by Hindman et al. was based mostly on cell morphology and size. (8) while studying an epidemic of pyrogenic re- Relatively very few comparative taxonomic actions among kidney dialysis patients. This ep- studies on physiological and chemical properties idemic occurred at the same time that maximum of cyanobacterial cells were published (23). It is, concentrations of algae and cyanobacteria were therefore, possible that the chemical character- observed in the raw water. ization of cyanobacterial LPS or other cell wall The oral ingestion of endotoxins produced by components may shed more light on the taxon- cyanobacterial blooms in the treated water omy of cyanobacteria. stored in uncovered reservoirs may be responsi- There is some evidence that not all the species ble for some of the water-borne disease out- of cyanobacteria contain LPS. Wang and Hill breaks presently classified as "of unknown etiol- (28) believe that the occurrence of LPS among ogy." For example, the presence of high concen- cyanobacteria cannot be generalized. They iso- trations of endotoxins (2,500 ng ml-') in the lated only a polysaccharide fromAnabaena flos- Sewickley, Pa., water system measured by the aquae, while the lipid A part was completely Environmental Protection Agency shortly after absent. However, Weckesser et al. (29) isolated the outbreak of water-borne gastroenteritis of a true LPS from another species of the same unknown origin may lead to a speculation on a genus (Anabaena variabilis). Such an interspe- possible relationship between the consumption cific difference in biochemical composition of of LPS derived from S. calcicola (400,000 cells procaryotic cells is rather surprising, especially ml-l) and the reported illness. Under normal when considering the yield of the fatty acid operating conditions during the spring and sum- material obtained by Weckesser et al. (29) from mer of 1977, concentrations of S. calcicola in acid hydrolysates by extraction with petroleum this reservoir were much lower, ranging from 0 ether (10.7% of the LPS dry weight). Therefore, to 1,200 cells ml-', and the corresponding LPS it is important to examine all the species of concentrations were between 25 and 250 ng ml-'. cyanobacteria commonly occurring in water sys- Peptidoglycans ofgram-positive bacteria were tems for LPS content. described as having endotoxin-like properties as The isolated endotoxin from S. calcicola was well but at much higher concentrations than nontoxic to mice when injected intraperitoneally LPS (20). Even though Stanier and Cohen-Ba- but caused LAL gelation and positive Schwartz- zire (23) have shown that cyanobacterial cells man reaction. The low toxicity of LPS to mice contain a peptidoglycan layer which varies in cannot be considered a decisive factor in deter- width with different species, our data indicate mining the health effects of this substance to that this constituent probably occurs in S. cal- humans. All LPS isolated from six species of cicola in small quantities and does not contrib- cyanobacteria have been found to be nontoxic to ute substantially to LAL reaction. Our recent mice. Even LPS isolated from Yersinia pestis work has shown that even a thorough treatment exert very low toxicity, and 5 to 10 mg of this ofS. calcicola cells by lysozyme did not decrease endotoxin is required to kill a mouse. Endotoxin the LAL gelation activity (lysozyme breaks isolated from R-mutants of gram-negative bac- down the peptidoglycans into inactive com- teria have also much lower toxicity to mice than pounds). LPS derived from wild types (16). ACKNOWLEDGMENT All types of bacterial LPS may cause endotox- The work was supported by an Environmental Protection emia, especially in debilitated, immunosup- Agency research grant (R805754010). VOL. 38, 1979 SCHIZOTHRIX CALCICOLA LIPOPOLYSACCHARIDES 477 We thank Ian Campbell and Dennis Doerfler for perform- 17. Mikheyskaya, L. V., R. G. Ovodova, and S. Yu. Ovo- ing gas-liquid chromatographic and mass spectrometric analy- dov. 1977. Isolation and characterization of lipopoly- ses of fatty acids in prepared samples. saccharides from cell walls of blue-green algae of the genus Phormidium. J. Bacteriol. 130:1-3. LITERATURE CITED 18. Moore, R. E. 1977. Toxins from blue-green algae. Biosci- 1. Banner, A. H. 1966. Marine toxins from the Pacific. I. ence 27:797-802. Advances in the investigation of fish toxins, p. 157-165. 19. Orskov, F., I. Orskov, K. Jann, and B. Jann. 1967. In F. E. Russel and P. R. Saunders (ed.), Animal toxins. The sugar compounds of Escherichia coli 0 antigens: Pergamon Press, New York. examination of 100 antigenic test strains. Acta Pathol. 2. Buttke, T. M., and L. 0. Ingram. 1975. Comparison of Microbiol. Scand. 71:339-358. lipopolysaccharides fromAgmenellum quadruplicatum 20. Rotta, J. 1975. Endotoxin-like properties of the peptido- to Escherichia coli and Salnonella typhimurium by glycan. Z. Immunitaetsforsch. 149:230-244. using thin-layer chromatography. J. Bacteriol. 124: 21. Sedlak, J., J. Keleti, and 0. Luederitz. 1970. Concern- 1566-1573. ing the taxonomy and immunochemistry of the genus 3. Dean, A. G., and T. C. Jones. 1972. Seasonal gastroen- Citrobacter, p. 7-10. In Proceedings of the Vth Inter- teritis and malabsorption at an American military base national Congress for Infectious Diseases, vol. 2. Infec- in the Philippines. Am. J. Epidemiol. 95:111-127. tious Diseases Society, Vienna. 4. DiLuzio, N. R., and T. J. Friedmann. 1973. Bacterial 22. Sedlak, J., M. Puchmayerova-Slajsova, J. Keleti, and endotoxins in the environment. Nature (London) 244: 0. Luederitz. 1971. On the taxonomy, ecology and 49-51. immunochemistry of genus Citrobacter. J. Hyg. Epi- 5. Drouet, F. 1963. Ecophenes of Schizothrix calcicola (Os- demiol. Microbiol. Immunol. 15:366-374. cillatoriaceae). Proc. Acad. Nat. Sci. Philadelphia 115: 23. Stanier, R. Y., and G. Cohen-Bazire. 1977. Photo- 261-281. trophic prokaryotes: the cyanobacteria. Annu. Rev. Mi- 6. Golecki, J. K. 1974. Zur Feinstruktur der Zellwand bei crobiol. 31:225-274. einigen Cyanophyceen. Zentralbl. Bakteriol. Parasi- 24. Suzuki, M., T. Mikanu, I. Matsumoto, and S. Suzuki. tenkd. Infektionskr. Hyg. Abt. 1 Orig. Reihe A 228:189- 1977. Gelation of Limulus lysate by synthetic dextran 192. derivatives. Microbiol. Immunol. 21:419-425. 7. Gupta, R. S., and M. S. Dashora. 1977. Algal pollutants 25. Tisdale, E. S. 1931. Epidemic of intestinal disorders in and potable water, p. 431-459. In R. B. Pajasek (ed.), Charleston, W.Va., occurring simultaneously with un- Drinking water quality enhancement through source precedented water supply conditions. Am. J. Public protection. Science Publishers, Ann Arbor, Mich. Health 21:198-200. 8. Hindman, S. H., M. S. Favero, and A. Peterson. 1975. 26. Veldee, M. V. 1931. Epidemiological study of suspected Pyrogenic reactions during haemodialysis caused by waterborne gastroenteritis. Am. J. Public Health 21: extramural endotoxins. Lancet 18:732-737. 1227-1235. 9. James, D. E. 1974. Culturing algae. Carolina Biological 27. Volk, W. V. 1968. Isolation of D-galacturonic acid 1- Supply Co., Burlington, N.C. phosphate from hydrolysates of cell wall lipopolysac- 10. Karkhanis, I. D., J. I. Zeltner, J. J. Jackson, and D. charide extracted from Xanthomonas comprestris. J. J. Carlo. 1978. A new improved microassay to deter- Bacteriol. 95:782-786. mine 2-keto-3-deoxyoctonate in lipopolysaccharide of 28. Wang, A. W., and A. Hill. 1977. Chemical analysis of gram-negative bacteria. Anal. Biochem. 85:595-601. the phenol water-extractable material from Anabaena 11. Kaufmann, F., 0. Luederitz, H. Stierlin, and 0. West- flos-aquae. J. Bacteriol. 130:558-570. phal. 1962. Zur Chemie der O-Antigene von Enterobac- 29. Weckesser, J., A. Katz, G. Drews, H. Mayer, and L. teriaceae. I. Analyse der Zuckerbausteine von Salmo- Fromme. 1974. Lipopolysaccharide containing L-acof- neUa-O-Antigenen. Zentralbl. Bakteriol. Parasitenkd. riose in the filamentous blue-green alga Anabaena var- Infektionskr. Hyg. Abt. 1 178:442-458. iabilis. J. Bacteriol. 120:672-678. 12. Keleti, G., and W. H. Lederer. 1974. Handbook of 30. Weise, G., G. Drews, B. Jann, and K. Jann. 1970. micromethods for the biological sciences. Van Nostrand Identification and analysis of a lipopolysaccharide in Reinhold Co., New York. cell walls of the blue-green alga Anacystis nidulans. 13. Keleti, J., 0. Luederitz, D. Mlynarcik, and J. Sedlak. Arch. Mikrobiol. 71:89-98. 1971. Immunochemical studies of Citrobacter 0 anti- 31. Westphal, O., and K. Jann. 1965. Bacterial lipopolysac- gens (lipopolysaccharides). Eur. J. Biochem. 20:237- charides, extraction with phenol water and further ap- 244. plications of the procedure. Methods Carbohydr. Chem. 14. Keleti, J., H. Mayer, I. Fromme, and 0. Luederitz. 5:83-91. 1970. The identification of 4-deoxy-D-arabino-hexose as 32. Wildfeuer, A., B. Heymer, K. H. Schleifer, H. P. a constituent in lipopolysaccharides of four Citrobacter Seidel, and 0. Haferkamp. 1974. Zur Schockdiagno- species. Eur. J. Biochem. 20:284-288. stik: der Nachweis von Endotoxin und Mucopeptid mit 15. Lippy, E. C., and J. Erb. 1976. Gastrointestinal illness at dem Limulus polyphemus-Lysat-Test. Klin. Wo- Sewickley, Pa. J. Am. Waterworks Assoc. 68:606-610. chenschr. 52:175-178. 16. Luederitz, O., K. Jann, and K. Wheat. 1967. Somatic 33. Wildfeuer, A., B. Heymer, D. Spilker, K. H. Schleifer, and capsular antigens of gram-negative bacteria, p. 60- E. Vanek, and 0. Haferkamp. 1975. Use of Limulus 183. In M. Florkin and E. H. Stotz (ed.), Comprehensive assay to compare the biological activity of peptidogly- biochemistry, vol. 26A. Elsevier, Amsterdam. can and endotoxin. Z. Immunitaetsforsch. 149:258-264.